MCP.so
ログイン
P

Promptheus

@abhichandra21

リポジトリ

Promptheus について

AI-powered prompt refinement tool with adaptive questioning and multi-provider support. Provides 5 MCP tools (refine_prompt, tweak_prompt, list_models, list_providers, validate_environment) for intelligent prompt engineering. Supports Google Gemini, Anthropic Claude, OpenAI, Groq

#prompt-engineering#ai#llm#mcp#multi-provider#gemini#claude#openai#groq#refinement#prompt-optimization

基本情報

カテゴリ

AI とエージェント

トランスポート

stdio

公開者

abhichandra21

投稿者

Abhishek Chandra

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "promptheus": {
      "type": "stdio",
      "command": "~/Promptheus/venv/bin/python",
      "args": [
        "-m",
        "promptheus.main",
        "mcp"
      ],
      "env": {
        "ANTHROPIC_API_KEY": "API-KEY"
      }
    }
  }
}

ツール

ツールは検出されませんでした

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is Promptheus?

Promptheus is a Model Context Protocol (MCP) server that analyzes, refines, and optimizes prompts for large language models. It integrates with multiple AI providers (Google, OpenAI, Anthropic, Groq, Qwen, etc.) and exposes its capabilities as standardized MCP tools for use by MCP-compatible clients.

How to use Promptheus?

Install with pip install promptheus and ensure at least one provider API key is set in a .env file or environment variables. Start the MCP server with promptheus mcp or python -m promptheus.mcp_server. The server provides tools such as refine_prompt, tweak_prompt, list_models, list_providers, and validate_environment. An interactive CLI mode (promptheus) and a web UI (promptheus web) are also available.

Key features of Promptheus

  • Adaptive task detection (refinement vs. direct optimization)
  • Multi-provider support (Google, OpenAI, Anthropic, Groq, Alibaba, Zhipu, OpenRouter)
  • Interactive refinement through natural conversation
  • Session history tracking and reuse
  • Structured clarification workflow with AskUserQuestion integration
  • CLI and web UI for prompt management
  • Telemetry and analytics (local, anonymous, can be disabled)

Use cases of Promptheus

  • Refine a vague prompt into a detailed, audience-aware request
  • Iteratively improve prompts through targeted clarification questions
  • Apply surgical modifications to existing prompts (e.g., “make it shorter”)
  • Discover and validate available models and provider configurations
  • Integrate prompt optimization into automated pipelines or MCP clients

FAQ from Promptheus

What providers are supported?

Google Gemini, Anthropic Claude, OpenAI, Groq, Alibaba Qwen, Zhipu GLM, and OpenRouter (optimized for openrouter/auto). Each requires its own API key.

How do I configure API keys?

Create a .env file with keys like GOOGLE_API_KEY=... or run promptheus auth for interactive setup. Use list_providers or validate_environment to check configuration.

What does the MCP server do exactly?

It exposes tools for prompt refinement (refine_prompt), prompt tweaking (tweak_prompt), model discovery (list_models), provider validation (list_providers), and environment testing (validate_environment). It supports a two‑phase workflow with optional clarification questions.

Are there any runtime dependencies?

Python 3.10+ and the mcp package (pip install mcp). The server connects to external LLM APIs; no local model is required.

Can I disable telemetry?

Yes. Set PROMPTHEUS_TELEMETRY_ENABLED=0 in your environment, or view a summary with promptheus telemetry summary. All data is stored locally.

コメント

「AI とエージェント」の他のコンテンツ

LinkedIn MCP Server

stickerdaniel

Open-source MCP server for LinkedIn. Give Claude and any MCP-compatible AI agent access to profiles, companies, jobs, and messages.

2.7K

1MCP - One MCP Server for All

1mcp-app

A unified Model Context Protocol server implementation that aggregates multiple MCP servers into one.

468

Open Multi-Agent Canvas

CopilotKit

The open-source multi-agent chat interface that lets you manage multiple agents in one dynamic conversation and add MCP servers for deep research

507

1Panel

1Panel-dev

🔥 1Panel is a modern, open-source VPS control panel — and the only one with native AI agent support. Run Ollama models, deploy OpenClaw agents, and manage your entire server stack from one clean web interface.

36.1K

Perplexity MCP Server

DaInfernalCoder

A Model Context Protocol (MCP) server for research and documentation assistance using Perplexity AI. Won 1st @ Cline Hackathon

295